Cloud Computing Use Cases
A white paper produced by the Cloud Computing Use Case Discussion Group
Version 4.0 2 July 2010 Contributors: Miha Ahronovitz, Dustin Amrhein, Patrick Anderson, Andrew de Andrade, Joe Armstrong, Ezhil Arasan B, James Bartlett, Richard Bruklis, Ken Cameron, Mark Carlson, Reuven Cohen, Tim M. Crawford, Vikas Deolaliker, Pete Downing, Andrew Easton, Rodrigo Flores, Gaston Fourcade, Thomas Freund, Tom Hanan, Valery Herrington, Babak Hosseinzadeh, Steve Hughes, William Jay Huie, Nguyen Quang Hung, Pam Isom, Shobha Rani J, Sam Johnston, Ravi Kulkarni, Anil Kunjunny, Edmond Lau, Thomas Lukasik, Bob Marcus, Gary Mazzaferro, Craig McClanahan, Meredith Medley, Walt Melo, Andres Monroy-Hernandez, Ayman Nassar, Dirk Nicol, Lisa Noon, Santosh Padhy, Gilad Parann-Nissany, Greg Pfister, Thomas Plunkett, Ling Qian, Balu Ramachandran, Jason Reed, German Retana, Bhaskar Prasad Rimal, Dave Russell, Matt F. Rutkowski, Clark Sanford, Krishna Sankar, Alfonso Olias Sanz, Mark B. Sigler, Wil Sinclair, Erik Sliman, Patrick Stingley, Phillip Straton, Robert Syputa, Robert J. Taylor, Doug Tidwell, Kris Walker, Kurt Williams, John M Willis, Yutaka Sasaki, Michael Vesace, Eric Windisch, Pavan Yara and Fred Zappert. Public comments on this document are welcomed and encouraged via the discussion groups referenced at http://cloudusecases.org.

This work is licensed under a Creative Commons Attribution Share Alike 3.0 Unported License.

Cloud Computing Use Cases White Paper

Version 4.0

Table of Contents
1 Introduction.........................................................................................................4 2 Definitions and Taxonomy..................................................................................6 2.1 Definitions of Cloud Computing Concepts...................................................6 2.2 Taxonomy...................................................................................................10 2.3 Relationships Between Standards and Taxonomies.................................13 2.4 Application Programming Interfaces (APIs)...............................................15 3 Use Case Scenarios.........................................................................................18 3.1 End User to Cloud......................................................................................19 3.2 Enterprise to Cloud to End User................................................................20 3.3 Enterprise to Cloud....................................................................................23 3.4 Enterprise to Cloud to Enterprise...............................................................24 3.5 Private Cloud..............................................................................................26 3.6 Changing Cloud Vendors...........................................................................27 3.7 Hybrid Cloud...............................................................................................29 3.8 Cross-Reference: Requirements and Use Cases......................................31 4 Customer Scenarios.........................................................................................33 4.1 Customer Scenario: Payroll Processing in the Cloud................................33 4.2 Customer Scenario: Logistics and Project Management in the Cloud......35 4.3 Customer Scenario: Central Government Services in the Cloud..............36 4.4 Customer Scenario: Local Government Services in a Hybrid Cloud.........37 4.5 Customer Scenario: Astronomic Data Processing.....................................38 5 Developer Requirements..................................................................................40 6 Security Scenarios............................................................................................43 6.1 Regulations ...............................................................................................43 6.2 Security Controls........................................................................................44 6.3 Security Federation Patterns ....................................................................46 7 Security Use Case Scenarios...........................................................................48 7.1 Computing Power in the Cloud..................................................................48 7.2 Cloud-based Development and Testing.....................................................49 7.3 Storage in the Cloud..................................................................................50 7.4 Cross-Reference: Security Controls and Customer Scenarios.................52

2 July 2010

2

...........1 What is an SLA? ...............61 8...... Service Level Agreements (SLAs)...............................54 8.................55 8.........4 Considerations for SLAs...............................................................................................................53 8 Service Level Agreements (SLAs)......56 8..................67 About Version 4: New to Version 4 is Section 8.......................8 Cross-reference: SLA Requirements and Cloud Delivery Models.........................56 8........................................................................................7 Cross-reference: SLA Requirements and Use Case Scenarios..................65 Summary of Changes................63 9 Conclusions and Recommendations....0 7...........................................................................................................5 SLA requirements...... 2 July 2010 3 .......................................................................................................58 8....................................................................62 8.............................Cloud Computing Use Cases White Paper Version 4..54 8................3 Service Level Management....6 A note about reliability.......................................................... See the Summary of Changes on page 67 for complete details.....2 Service Level Objectives..............5 Cross-Reference: Security Federation Patterns and Customer Scenarios............

Cloud Computing Use Cases White Paper Version 4. The goal of this white paper is to highlight the capabilities and requirements that need to be standardized in a cloud environment to ensure interoperability.  Make it clear where existing standards should be used. or if it can only be built with proprietary APIs and products. The IT industry has invested heavily in existing standards and standards organizations. The use cases:  Provide a practical. If a particular use case can't be built today. the industry needs to define standards to make that use case possible. there is no need to duplicate or reinvent them. 250 organizations signed on as supporters. proprietary technologies. A use case that clearly describes a common task and outlines the difficulties in accomplishing it is the best possible justification for any standards effort. 2 July 2010 4 .0 1 Introduction The Cloud Computing Use Case group brought together cloud consumers and cloud vendors to define common use case scenarios for cloud computing. customer-experience-based context for discussions on interoperability and standards. This group's activity is done in light of the six principles of the Open Cloud Manifesto:  Cloud providers must work together to ensure that the challenges to cloud adoption are addressed through open collaboration and the appropriate use of standards. Cloud computing must evolve as an open environment.org) is a statement of the principles for maintaining openness in cloud computing.  Focus the industry's attention on the importance of Open Cloud Computing.  Make it clear where there is standards work to be done.  Cloud providers must use and adopt existing standards wherever appropriate. ease of integration and portability. Within two months of its announcement. The Open Cloud Manifesto (opencloudmanifesto. minimizing vendor lock-in and increasing customer choice. The use case scenarios demonstrate the performance and economic benefits of cloud computing and are based on the needs of the widest possible range of consumers. It must be possible to implement all of the use cases described in this paper without using closed.

2 July 2010 5 .  Cloud computing standards organizations. This paper is part of the ongoing effort to make these principles a reality.Cloud Computing Use Cases White Paper Version 4. making sure that efforts do not conflict or overlap. we must be judicious and pragmatic to avoid creating too many standards.  Cloud providers must not use their market position to lock customers into their particular platforms and limiting their choice of providers. We must ensure that standards promote innovation and do not inhibit it.  Any community effort around the open cloud should be driven by customer needs. not merely the technical needs of cloud providers.0  When new standards (or adjustments to existing standards) are needed. advocacy groups. and communities should work together and stay coordinated. and should be tested or verified against real customer requirements.

 Infrastructure as a Service (IaaS): The consumer uses "fundamental computing resources" such as processing power.  Platform as a Service (PaaS): The consumer uses a hosting environment for their applications. and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.gov/groups/SNS/cloud-computing/. but not the cloud infrastructure beneath them.S. storage.Cloud Computing Use Cases White Paper Version 4. regardless of how those scenarios might be defined or placed into a taxonomy. (This definition is from the latest draft of the NIST Working Definition of Cloud Computing published by the U. hardware or network infrastructure on which they are running. 2. The document states. networks. The consumer can control the operating system. applications. delivery models and deployment models discussed in this paper are based on Version 15 of the document. not defining cloud computing itself. The consumer controls the applications that run in the environment (and possibly has some control over the hosting environment). on-demand network access to a shared pool of configurable computing resources (e. interesting and useful. Our goal is to provide use case scenarios that are clear.0 2 Definitions and Taxonomy The following definitions and taxonomy are included to provide an overview of cloud computing concepts. Government's National Institute of Standards and Technology.. 2 July 2010 6 . hardware or network infrastructure on which it's running. deployed applications and possibly networking components such as firewalls and load balancers.g. the focus of this white paper is defining cloud scenarios and use cases based on real-world applications and requirements.1 Delivery Models The NIST definition of cloud computing defines three delivery models:  Software as a Service (SaaS): The consumer uses an application. However.1) 2. but does not control the operating system. networking components or middleware. 1 You can find the full document on the NIST Cloud Computing page at http://csrc. but does not control the operating system.nist." The essential characteristics. convenient. dated 8-19-09.1. storage.1 Definitions of Cloud Computing Concepts Cloud Computing: Cloud computing is a model for enabling ubiquitous. "This material is public domain although attribution to NIST is requested. servers. It may be freely duplicated and translated. The platform is typically an application framework. storage.

The term “public” does not always mean free. such as specific security requirements or a common mission.2 Deployment Models The NIST definition defines four deployment models:  Public Cloud: In simple terms.  Private Cloud: A private cloud offers many of the benefits of a public cloud computing environment.  Hybrid Cloud: A hybrid cloud is a combination of a public and private cloud that interoperates. To the consumer.2  Community Cloud: A community cloud is controlled and used by a group of organizations that have shared interests.7. security exposures and legal requirements that using public cloud services might entail.3 2. and the consumer can purchase as much or as little computing power as they need.3 Essential Characteristics The NIST definition describes five essential characteristics of cloud computing. For that reason. improving security and resiliency because user access and the networks used are restricted and designated. The members of the community share access to the data and applications in the cloud. In addition.1. while keeping business-critical services and data in their control.0 2. A private cloud can be managed by a third party and can be physically located off premises.Cloud Computing Use Cases White Paper Version 4. A public cloud does not mean that a user’s data is publically visible. It is not necessarily managed and hosted by the organization that uses it. the cloud appears to be infinite. public cloud vendors typically provide an access control mechanism for their users.1. 3 2 July 2010 7 . even though it can be free or fairly inexpensive to use. 2 A Hybrid Cloud is a superset of the technology used in a Community Cloud.  Rapid Elasticity: Elasticity is defined as the ability to scale resources both up and down as needed. private cloud services offer the provider and the user greater control of the cloud infrastructure. In this model users typically outsource non-businesscritical information and processing to the public cloud. cost effective means to deploy solutions. The difference between a private cloud and a public cloud is that in a private cloud-based service. the requirements for the two deployment models are discussed together under the heading "Hybrid Cloud" in Section 3. such as being elastic and service based. Public clouds provide an elastic. data and processes are managed within the organization without the restrictions of network bandwidth. public cloud services are characterized as being available to clients from a third party service provider via the Internet. This is one of the essential characteristics of cloud computing in the NIST definition.

aspects of the cloud service are controlled and monitored by the cloud provider.0  Measured Service: In a measured service.4  Resource Pooling: Resource pooling allows a cloud provider to serve its consumers via a multi-tenant model.Cloud Computing Use Cases White Paper Version 4. or datacenter).5 2.4 Other Terms Interoperability: Interoperability is concerned with the ability of systems to communicate.  On-Demand Self-Service: The on-demand and self-service aspects of cloud computing mean that a consumer can use cloud services as needed without any human interaction with the cloud provider. this means the ability to write code that works with more than one cloud provider simultaneously.g. state. 2 July 2010 8 . Service Level Agreement (SLA): An SLA is contract between a provider and a consumer that specifies consumer requirements and the provider’s commitment 4 This does not necessarily mean Internet access.6 Portability: Portability is the ability to run components or systems written for one environment in another environment. country.testingstandards. It requires that the communicated information is understood by the receiving system. this includes software and hardware environments (both physical and virtual).  Ubiquitous Network Access: Ubiquitous network access means that the cloud provider’s capabilities are available over the network and can be accessed through standard mechanisms by both thick and thin clients. access to the cloud is typically not limited to a particular type of client. Integration among cloud-based components and systems can be complicated by issues such as multi-tenancy. The cloud provider and cloud consumer must work together to adhere to those regulations. Physical and virtual resources are assigned and reassigned according to consumer demand.htm. This is crucial for billing. There is a sense of location independence in that the customer generally has no control or knowledge over the exact location of the provided resources but may be able to specify location at a higher level of abstraction (e. By definition. resource optimization. 6 The definitions of interoperability. Regardless of the type of network. Integration: Integration is the process of combining components or systems into an overall system. a private cloud is accessible only behind a firewall. In the world of cloud computing. capacity planning and other tasks. portability and integration are based on the work at http://www..uk/interop_et_al.1. federation and government regulations. In the world of cloud computing.co. 5 In many cases privacy laws and other regulations require the cloud provider's resources to be in a particular location. regardless of the differences between the providers. access control.

 APIs can be defined in specific programming languages or in more neutral formats such as WSDL or IDL. Application Programming Interface (API): An application programming interface is a contract that tells a developer how to write code to interact with some kind of system. the hybrid cloud automatically allocates additional resources to the private cloud. the API specifies the information that should be sent to the system. Changes made to the VM while it is running can be stored to disk to make them persistent. Federation can be done by a cloud provider or by a cloud broker. The consumer has no knowledge that the broker does not control the resources. applications or data from different enterprises hosted on the same physical hardware. 2 July 2010 9 . The API describes the syntax of the operations supported by the system. looks to the user like an actual machine. Virtual Machine (VM): A file (typically called an image) that. Multitenancy is common to most cloud-based systems. Broker: A broker has no cloud resources of its own. Federation: Federation is the act of combining data or identities across multiple systems.0 to them. Infrastructure as a Service is often provided as a VM image that can be started or stopped as needed. Typically an SLA includes items such as uptime. Governance: Governance refers to the controls and processes that make sure policies are enforced. the hybrid cloud is not used. the information that the system will send back. REST specifications typically don't have a machine-readable language. a security policy might specify that all requests to a particular cloud service must be encrypted. privacy. Policy: A policy is a general term for an operating procedure. but matches consumers and providers based on the SLA required by the consumer. Cloud bursting: Cloud bursting is a technique used by hybrid clouds to provide additional resources to private clouds on an as-needed basis. and any error conditions that might occur. Multi-Tenancy: Multi-tenancy is the property of multiple systems. when executed.Cloud Computing Use Cases White Paper Version 4. When workloads exceed the private cloud’s capacity. For example. but they define an API nonetheless. If the private cloud has the processing power to handle its workloads. security and backup procedures.  An API can also include the details of protocols (such as HTTP) and data formats (such as JSON or an XML Schema). For each operation.

2 July 2010 10 .Cloud Computing Use Cases White Paper Version 4.0  An API requires human intelligence to understand the semantics of the data and operations.2 Taxonomy This diagram defines a taxonomy for cloud computing: In this diagram. a human being. but a developer. (Notice that open standards are needed for the interactions between these roles. 2. has to figure out which of infinity's two integers should be used.) Each role is discussed in more detail in the following sections. Service Providers manage the cloud infrastructure and Service Developers create the services themselves. Service Consumers use the services provided through the cloud. A machine can discover that method x requires two integers as its parameters.

Other user interfaces provide administrative functions such as starting and stopping virtual machines or managing cloud storage. Consumers writing application code use different programming interfaces depending on the application they are writing.1 Service Consumer The service consumer is the end user or enterprise that actually uses the service. 2.2. Consumers work with SLAs and contracts as well. Platform or Infrastructure as a Service.2 Service Provider 2 July 2010 11 . The expectations of the consumer and the reputation of the provider are a key part of those negotiations. Typically these are negotiated via human intervention between the consumer and the provider.Cloud Computing Use Cases White Paper Version 4. Depending on the type of service and their role. the consumer does not need to know about cloud computing as they use the application. Some user interfaces look like any other application. whether it is Software.0 2. the consumer works with different user interfaces and programming interfaces.2.

the lowest layer of the stack is the firmware and hardware on which everything else is based. the provider manages the cloud infrastructure for the platform. capacity planning to ensure that consumer demands will be met. The consumer uses that service as if it were a disk drive. The virtual images controlled by the VM manager include both the images themselves and the metadata required to manage them. message queue or other middleware.) Open standards apply to the provider’s operations as well. SLA management to ensure that the terms of service agreed to by the provider and consumer are adhered to. either the operating system or virtual machine manager that hosts the infrastructure beneath the cloud. typically a framework for a particular type of application. or machine. the provider installs. The actual task of the provider varies depending on the type of service:  For Software as a Service. the consumer does not have access to the infrastructure. Above that is the software kernel. the provider maintains the storage. In the service provider diagram. 2 July 2010 12 .Cloud Computing Use Cases White Paper Version 4. (The many levels of security requirements are beyond the scope of this paper. management involves billing to recover costs. and monitoring to track the status of the system and its resources.  For Platform as a Service. The virtualized resources and images include the basic cloud computing services such as processing power. message queue.  For Infrastructure as a Service. At a low level. and reporting for administrators. or the hosting environment for virtual machines. database. At a higher level. The provider does not necessarily own the physical infrastructure in which the software is running. A well-rounded set of standards simplify operations within the provider and interoperability with other providers. The consumer’s application cannot access the infrastructure underneath the platform.0 The service provider delivers the service to the consumer. but they cannot access the infrastructure that hosts it. storage and middleware. Security applies to all aspects of the service provider’s operations. database. manages and maintains the software. management requires metering to determine who uses the services and to what extent. Regardless. Crucial to the service provider’s operations is the management layer. provisioning to determine how resources are allocated to consumers. they can access only the application.

they are most likely writing code for an environment hosted by a cloud provider.0 2. 2. publishing the service is deploying it to the cloud provider’s infrastructure. In this case. across the different types of cloud services. Standards will have an impact within each type of cloud service. publishes and monitors the cloud service. These are typically "line-of-business" applications that are delivered directly to end users via the SaaS model. If developers are creating a SaaS application. between the enterprise and the cloud. and within the private cloud of an enterprise. analytics involve remote debugging to test the service before it is published to consumers.Cloud Computing Use Cases White Paper Version 4.2. During service creation.3 Service Developer The service developer creates. analytics allow developers to monitor the performance of their service and make changes as necessary. 2 July 2010 13 . Development environments for service creation vary. Once the service is published.3 Relationships Between Standards and Taxonomies There are four different ways standards will affect cloud use case scenarios. Applications written at the IaaS and PaaS levels will subsequently be used by SaaS developers and cloud providers.

a cloud message queue. a cloud-based word processing application should support standards for document portability. the requirement for standards support in a word processing application has nothing to do with whether the application is running in the cloud. open standards apply at the application level.0 2. For Platform as a Service. Standards that define how an enterprise application communicates with resources such as a cloud database or a cloud message queue would enable those applications to use cloud services with little or no changes. Very little of the standards work here is cloud-specific. a standard set of APIs to work with cloud databases would allow applications to work with data from multiple vendors. so those standards are beyond the scope of this paper. For example.1 Standards Across Cloud Service Types As cloud computing becomes more common. and manage (start/stop/monitor) virtual machines running in the cloud.Cloud Computing Use Cases White Paper Version 4. Standards to define how these different services work together should provide value. That common API would give users the freedom to move to another cloud database provider without major changes. For Infrastructure as a Service.3 Standards Between the Cloud and the Enterprise Even as cloud computing emerges. open standards make it possible to avoid vendor lock-in. as would common formats for data and data interchange. An application might use a cloud storage service. enterprise architectures such as Java EE are not going away. Figuring out how to integrate cloud computing with existing architectures and development paradigms will be a major challenge for this group. 2.2 Standards Within Cloud Service Types Within each type of cloud service (IaaS. In the case of virtual machines.3. applications will likely use different types of cloud services. message queues or MapReduce would provide similar benefits. but they are accessible only through the APIs of the framework. 2. Common APIs for other cloud infrastructure services such as storage.3. many of the platforms provided in the cloud are application frameworks. 2 July 2010 14 . PaaS or SaaS). storage and databases. and it would make it much easier to integrate new data sources with existing applications. a common virtual machine format is crucial. Users should be able to take a VM built and deployed with one cloud provider and deploy it to another cloud provider without changes.3. For Software as a Service. Those frameworks typically provide common services such as user interfaces.

and will build upon the standards that apply between enterprises and the cloud. If the service is REST-based. for example) are handled by the toolkit.0 2. Although developers are still focused on the format and structure of the data going across the wire. 2. many of the details (handling response codes and calculating signatures.4. adds the appropriate SOAP headers. The REST service returns data with an accompanying HTTP response code. security and management. The enterprise will interact with some combination of private. Level 1 – The Wire: At this level. and fills the body of the SOAP envelope with the data payload. A developer working at this level uses a common interface to multiple cloud computing 15 2 July 2010 . and they fall into five basic categories. creates the payload for the request. Working at this level.3. the developer creates the SOAP envelope. auditability. 2.4 Application Programming Interfaces (APIs) The primary mechanism for building cloud computing solutions is the APIs provided by cloud providers.4 Standards Within an Enterprise Standards within an enterprise will be determined by requirements such as interoperability. for that reason. Level 3 – Service-Specific Toolkits: The developer uses a higher-level toolkit to work with a particular service. and opens an HTTP connection to the service. Each level requires the developer to focus on different tasks and data structures. Because of the straightforward nature of many REST services.Cloud Computing Use Cases White Paper Version 4. the developer is able to focus on business objects and business processes. it is possible to be relatively efficient while writing code at this level. If the service is SOAP-based. Cloud APIs work at four different levels. the developer writes directly to the wire format of the request. Working with SOAP services requires parsing the XML content of the envelopes. public and hybrid clouds. most SOAP services are invoked with a higher-level API. A developer can be far more productive when focusing on the data and processes that matter to the organization instead of focusing on the wire protocol. The SOAP service responds with a SOAP envelope that contains the results of the request. Level 2 – Language-Specific Toolkits: Developers at this level use a languagespecific toolkit to work with SOAP or REST requests.1 Levels of APIs There are four different levels of APIs. the developer creates the appropriate HTTP headers. Level 4 – Service-Neutral Toolkits: This is the highest level of API.

it is helpful to point out the different roles that developers play. Java. 2. 2 July 2010 16 .Cloud Computing Use Cases White Paper Version 4. stopping.0 providers. restarting. An application written with a service-neutral toolkit should be able to use a different cloud vendor (see Changing Cloud Vendors on page 27) with minimal changes to the code. Infrastructure management APIs control details such as firewalls. along with the categories of APIs they use: Client Application developer: Writes cloud-based client applications for end users. PHP. In addition to any cloud-specific packaging technologies. the developer focuses on business objects and business processes. There is nothing cloud-specific in this category. this includes traditional packaging mechanisms such as .Net assemblies and EAR/WAR files.2 Categories of APIs Programming interfaces can be divided into five categories: Category 1 – Ordinary Programming: The usual application programming interfaces in C#. 2. Requirements for APIs and cloud services vary from one role to the next. These APIs are divided into subcategories for cloud storage services. As discussed in the previous section. and deleting images. These developers use APIs for cloud services (category 3). Category 4 – Image and Infrastructure Management: Programming interfaces to manage virtual machine images and infrastructure details. Here are the roles we will discuss in this document. cloud databases. if any.4. Category 5 – Internal Interfaces: Programming interfaces for the internal interfaces between the different parts of a cloud infrastructure. As with Level 3. a developer working at Level 4 does not have to worry about which cloud service they are working with. node management.3 Developer Roles To discuss requirements for developers. A developer writing code using cloud services APIs is aware that they are using the cloud. Category 3 – Cloud Services: Programming interfaces that work with cloud services. APIs for images support uploading.4. Category 2 – Deployment: Programming interfaces to deploy applications to the cloud. network management and load balancing. etc. These are the APIs you would use if you wanted to change vendors for the storage layer in your cloud architecture. and other cloud services. cloud message queues. cloud service APIs can be either service-specific or service-neutral. Unlike Level 3. deploying starting.

Deployers: Package. 2 July 2010 17 . cloud services. Lifecycle management is a concern here as well. including deployment and infrastructure management. These developers use APIs from category 5. Cloud Providers: Work with the infrastructure beneath their cloud offerings. These developers use ordinary APIs (category 1) as well as APIs for cloud services (category 3).Cloud Computing Use Cases White Paper Version 4.0 Application developer: Writes traditional applications that use the cloud. deploy and maintain applications that use the cloud. and 4. These developers use APIs from categories 2. and image management (categories 2. 3 and 4). 3. Administrators: Work with applications at multiple levels. These developers use APIs for deployment.

The graphics in this section have common elements throughout. so only the Enterprise appears in color. End User to Cloud Applications running on the cloud and accessed by end users Enterprise to Cloud to End User Applications running in the public cloud and accessed by employees and customers Enterprise to Cloud Cloud applications integrated with internal IT capabilities 2 July 2010 18 . As an example.Cloud Computing Use Cases White Paper Version 4. it is grayed out or drawn with a dashed line. If a given element does not apply to a particular use case.0 3 Use Case Scenarios The Enterprise Cloud Usage scenarios are intended to illustrate the most typical cloud use cases and are not meant to be an exhaustive list of realizations within a cloud environment. the Private Cloud use case does not involve the End User or the Public Cloud.

Hybrid Cloud Multiple clouds work together. user identity. their data is stored and managed in the cloud. security and other details. an end user is accessing data or applications in the cloud. The user doesn’t want to keep up with anything more than a password.1 End User to Cloud In this scenario. coordinated by a cloud broker that federates data. A user of Gmail. 3.Cloud Computing Use Cases White Paper Version 4. 2 July 2010 19 . applications. Common applications of this type include email hosting and social networking sites. Facebook or LinkedIn accesses the application and their data through any browser on any device. Changing Cloud Vendors An organization using cloud services decides to switch cloud providers or work with additional providers.0 Enterprise to Cloud to Enterprise Cloud applications running in the public cloud and interoperating with partner applications (supply chain) Private Cloud A cloud hosted by an organization inside that organization’s firewall.

the user has no idea how the underlying architecture works. an enterprise is using the cloud to deliver data and services to the end user.1. the enterprise accesses the cloud to retrieve data and / or manipulate it. A full discussion of security in cloud computing is beyond the scope of this paper. they can get to their data.  An open client: Access to the cloud service should not require a particular platform or technology. sending the results to 2 July 2010 20 .0 Most importantly. If they can get to the Internet.Cloud Computing Use Cases White Paper Version 4.  Security: Security (including privacy) is a common requirement to all use cases.  SLAs: Although service level agreements for end users will usually be much simpler than those for enterprises. although the details of those requirements will vary widely from one use case to the next. When the end user interacts with the enterprise. cloud vendors must be clear about what guarantees of service they provide.1 Requirements  Identity: The cloud service must authenticate the end user. 3.2 Enterprise to Cloud to End User In this scenario. 3.

 Federated identity: In addition to the basic identity needed by an end user. with an infrastructure federating other identities that might be required by cloud services. there might be legal restrictions on the location of the physical server where the data is stored. Many applications cannot be moved to the cloud until cloud vendors provide an API for determining the location of the physical hardware that delivers the cloud service. The ideal is that the enterprise user manages a single ID.  Location awareness: Depending on the kind of data the enterprise is managing on the user's behalf.Cloud Computing Use Cases White Paper Version 4. 2 July 2010 21 .2.1 Requirements  Identity: The cloud service must authenticate the end user. The end user can be someone within the enterprise or an external customer.0 the end user. an enterprise user is likely to have an identity with the enterprise. this requirement is essential.  An open client: Access to the cloud service should not require a particular platform or technology. Although this violates the cloud computing ideal that the user should not have to know details of the physical infrastructure. 3.

Other governance requirements will be industry. and there must be an unambiguous way of measuring what was actually delivered.  Data and Application Federation: Enterprise applications need to combine data from multiple cloud-based sources. Any solution to this requirement must account for differences in the ways cloud vendors attach storage to virtual machines. that ease of use creates the risk that individuals in an enterprise will use cloud services on their own initiative. chargebacks and provisioning.Cloud Computing Use Cases White Paper Version 4.)  Lifecycle Management: Enterprises must be able to manage the lifecycle of applications and documents. Management of VMs and of cloud services such as storage.  Common APIs for Cloud Storage and Middleware: The enterprise use cases require common APIs for access to cloud storage services. and other cloud middleware services such as message queues. enterprises who sign contracts based on SLAs will need a standard way of benchmarking performance. Writing custom code that works only for a particular vendor’s cloud service locks the enterprise into that vendor’s system and eliminates some of the financial benefits and flexibility that cloud computing provides. see the discussions of SLAs. Similarly. (SLAs have additional implications for cloud security.  Security: Any use case involving an enterprise will have more sophisticated security requirements than one involving a single end user.  SLAs and Benchmarks: In addition to the basic SLAs required by end users.0  Metering and monitoring: All cloud services must be metered and monitored for cost control. and they need to coordinate the activities of applications running in different clouds. This requirement includes versioning of 2 July 2010 22 .  Management and Governance: Public cloud providers make it very easy to open an account and begin using cloud services. the more advanced enterprise use cases to follow will have equally more advanced security requirements. databases and message queues is needed to track what services are used.  A Common File Format for VMs: A VM created for one cloud vendor’s platform should be portable to another vendor’s platform. There must be an unambiguous way of defining what a cloud provider will deliver. auditing and monitoring in Section 6 for more information. Governance is crucial to ensure that policies and government regulations are followed wherever cloud computing is used.and geography-specific. cloud databases.

shutting down those VMs when they're not needed anymore)  Using applications in the cloud (SaaS) for certain enterprise functions (email. CRM. calendaring. the enterprise uses cloud services to supplement the resources it needs:  Using cloud storage for backups or storage of seldom-used data  Using virtual machines in the cloud to bring additional processors online to handle peak loads (and. etc.3 Enterprise to Cloud This use case involves an enterprise using cloud services for its internal processes. This might be the most common use case in the early stages of cloud computing because it gives the enterprise the most control.0 applications and the retention and destruction of data. In this scenario. of course. In addition to data retention. 3. There are substantial legal liabilities if certain data is no longer available. 2 July 2010 23 .). in some cases an enterprise will want to make sure data is destroyed at some point.Cloud Computing Use Cases White Paper Version 4. Discovery is a major issue for many organizations.

location awareness.Cloud Computing Use Cases White Paper Version 4.0  Using cloud databases as part of an application's processing. metering and monitoring. Deployment of applications to the cloud should be straightforward as well. common APIs for cloud storage and middleware. it should be possible to move that image from one cloud provider to another. compensating for the different mechanisms vendors have for attaching storage to VMs. 3. management and governance. The applicable standards will vary from one application to the next and from one industry to the next. An open client. security. When that VM image is built. 3.  Industry-specific standards and protocols: Many cloud computing solutions between enterprises will use existing standards such as RosettaNet or OAGIS.4 Enterprise to Cloud to Enterprise This use case involves two enterprises using the same cloud.1 Requirements The basic requirements of the Enterprise to Cloud use case are much the same as those for the Enterprise to Cloud to End User use case. This could be extremely useful for sharing that database with partners.3. federated identity. Other requirements for this use case are:  Deployment: It should be simple to build a VM image and deploy it to the cloud as necessary. A supply chain is the most obvious example for this use case. The focus here is hosting resources in the cloud so that applications from the enterprises can interoperate. data and application federation. 2 July 2010 24 . etc. SLAs and lifecycle management all apply. a common file format for VMs. government agencies.

Identity. industry-specific standards. VM. data and application federation.0 3.Cloud Computing Use Cases White Paper Version 4. Other requirements for this use case are:  Transactions and concurrency: For applications and data shared by different enterprises. If two enterprises are using the same cloud-hosted application.1 Requirements The basic requirements of the Enterprise to Cloud to Enterprise use case are much the same as those for the Enterprise to Cloud use case. federated identity. location awareness. middleware 2 July 2010 25 . transactions and concurrency are vital. common APIs for storage and middleware. management and governance. metering and monitoring. SLAs and lifecycle management all apply.4. security. an open client.

With a private cloud. interoperability between the enterprises is essential. The payroll department gets extra cycles when they need it and other departments get extra cycles when they need it.0 or storage. deployment.5.1 Requirements The basic requirements of the Private Cloud use case are an open client. they need enough computing power to handle the maximum workload.  Interoperability: Because more than one enterprise is involved. even though their everyday workload for the rest of the month is much lower. metering and monitoring. management and governance. 3.Cloud Computing Use Cases White Paper Version 4. For example. and SLAs. if the payroll department has a surge in workload on the 15th and 30th of each month.5 Private Cloud The Private Cloud use case is different from the others in that the cloud is contained within the enterprise. it's important that any changes made by either enterprise are done reliably. computing power is spread across the enterprise. This can deliver significant savings across the enterprise. This is useful for larger enterprises. interoperability. 2 July 2010 26 . a common VM format. 3. security.

a common file format for VMs and common APIs for cloud storage and middleware. either adding an additional vendor or replacing an existing one. Keeping the cloud inside the enterprise removes many of the requirements for identity management. There are four different scenarios here. changing cloud vendors requires an open client. federated identity. Being able to work with other vendors without major changes is one of the main benefits of openness and standardization. industry standards.6 Changing Cloud Vendors This use case involves working with a different cloud vendor. location awareness. In many cases. transactions. security. SLAs. common APIs for cloud middleware and lifecycle management. location awareness. consumers have to use a private cloud so that location awareness will no longer be an issue.0 Note that a private cloud does not require identity. It applies to all of the other use cases discussed in this paper. 2 July 2010 27 . standards and common APIs. each of which has slightly different requirements. In general. 3.Cloud Computing Use Cases White Paper Version 4. The details of those requirements are discussed in each of the following subsections.

cloud middleware (databases. creating and dropping databases and tables.6. APIs for connecting to.6. etc. In some cases. The formats involved will depend on the type of application.).0 3. message queues and applications must be exportable from one vendor and importable by the other.  Common APIs for Cloud Middleware: This includes all of the operations supported by today’s cloud services. message queues.Cloud Computing Use Cases White Paper Version 4.1 Requirements  Industry-specific standards: Moving documents and data from one vendor’s application to another requires both applications to support common formats. cloud message queues and other middleware.1 Requirements  Industry-specific standards: Moving documents and data from one vendor’s middleware to another requires both applications to support common formats. 3. some cloud databases don't allow joins across tables. The formats involved will depend on the type of application. It is important to note that there is nothing cloud-specific to these requirements. The standards for moving a document from Zoho to Google Docs are the same standards for moving a document from Microsoft Office to OpenOffice. Map Reduce) and cloud storage are considered separate scenarios.6. In some cases. Existing data.6. Cloud database vendors have enforced certain restrictions to make their products more elastic and to limit the possibility of queries against large data sets taking significant resources to process. queries. and some don't support a true Because of the popularity of cloud storage. Documents and data created with one vendor's software should be importable by the second vendor's software. standard APIs for different application types will also be required. word processing. even though both are classified as PaaS.2 Scenario 2: Changing middleware vendors In this scenario a cloud customer changes cloud middleware vendors. including cloud databases.1 Scenario 1: Changing SaaS vendors In this scenario a cloud customer changes SaaS vendors. For example.1. 3. Both SaaS vendors provide the same application (CRM. 7 2 July 2010 28 .2. accounting.7 3. the customer might need to use the two vendors interchangeably.

Other middleware services such as message queues are more similar. 3. especially for applications built on a true relational model. 3. The provider of the hybrid cloud must manage cloud resources based on the consumer’s terms. the format of the URL and the driver name are different for different database vendors.3 Scenario 3: Changing cloud storage vendors In this scenario a cloud customer changes cloud storage vendors.0 database schema. 2 July 2010 29 .3.6. so there are no cloud-specific requirements for the software running inside the VM.6. as an example.Cloud Computing Use Cases White Paper Version 4. 3.1 Requirements  A common format for virtual machines: The VM format should work with any operating system. so finding common ground among them should be simpler.6. This means that the user of the virtual machine has chosen a platform prior to building a VM for the cloud. but the code to interact with the database is identical. including both public and private clouds.1 Requirements  A common API for Cloud Storage: Code that reads or writes data in one cloud storage system should work with a different system with as few changes as possible.7 Hybrid Cloud This use case involves multiple clouds working together. A broker can also deliver a hybrid cloud. A hybrid cloud can be delivered by a federated cloud provider that combines its own resources with those of other providers. In a JDBC application. The assumption here is that the virtual machines themselves are running an operating system such as Windows or Linux.6.4 Scenario 4: Changing VM hosts In this scenario a cloud customer wants to take virtual machines built on one cloud vendor's system and run it on another cloud vendor's system. those changes should be confined to configuration code.4. the difference is that a broker does not have any cloud resources of its own. 3. Those restrictions are a major challenge to moving between cloud database vendors. 3.

 SLAs: A machine readable.7.1 Requirements  All of the requirements of the previous use cases (except Transactions and concurrency) apply here. Data and Application Federation and Interoperability.1. A community cloud has an infrastructure shared among enterprises with a common purpose. This allows the hybrid cloud provider to select resources according to the consumer’s terms without human intervention. Here is the diagram for a community cloud: 2 July 2010 30 . particularly Security. the requirements for a community cloud are a subset of the requirements for the hybrid cloud. this use case is no different from the End User to Cloud use case discussed earlier.Cloud Computing Use Cases White Paper Version 4. standard format for expressing an SLA.0 It is important to note that to the consumer of a hybrid cloud.2. As mentioned in Section 2. The user has no knowledge of what the hybrid cloud provider actually does. 3.

This could be a VPN. but access is not via the public Internet.Cloud Computing Use Cases White Paper Version 4.8 Cross-Reference: Requirements and Use Cases The following table summarizes the relationships between requirements and use cases: Changing Cloud Vendors Private Cloud Enterprise to Cloud to End User Enterprise to Cloud Enterprise to Cloud to Enterprise Hybrid Cloud Requirement Identity Open Client Federated Identity Location Awareness Metering and Monitoring End User to Cloud                          2 July 2010 31 . 3.0 Notice that the communication between the community and the community cloud is done across an intranet.

0 Changing Cloud Vendors Private Cloud Enterprise to Cloud to End User Enterprise to Cloud Requirement Management and Governance Security Deployment Transactions and Concurrency Interoperability IndustrySpecific Standards VM Image Format Cloud Storage API Cloud Database API Cloud Middleware API Data and Application Federation SLAs Lifecycle Management       Enterprise to Cloud to Enterprise                                                    2 July 2010 32 Hybrid Cloud End User to Cloud .Cloud Computing Use Cases White Paper Version 4.

cloud storage Enterprise to Cloud Logistics & Project Management PaaS (app framework).1 Applicable Use Case Scenario from Section 3: Enterprise to Cloud 2 July 2010 33 . PaaS Private Cloud Local Government IaaS. cloud storage Enterprise to Cloud to End User 4. Here is a summary of those customer scenarios: Customer Scenario Customer Problem Solved  Processing time reduced  Hardware requirements reduced  Elasticity enabled for future expansion  Processing time reduced  Manual tasks eliminated  Development environment updated and streamlined  IT expertise consolidated  Hardware requirements reduced  IT expertise consolidated  Hardware requirements reduced  Hardware expense greatly reduced (processing power and storage)  Energy costs greatly reduced  Administration simplified Requirements & Capabilities Applicable Use Case Payroll Processing IaaS (VMs).Cloud Computing Use Cases White Paper Version 4.1 Customer Scenario: Payroll Processing in the Cloud 4.0 4 Customer Scenarios This section describes customer experiences with the use cases.1. PaaS Hybrid Cloud Astronomic Data Processing IaaS (VMs). cloud storage Enterprise to Cloud to End User Central Government IaaS.

two servers formerly dedicated to processing payroll were freed up for other tasks.1.0 4.5. a VM with a database server was deployed. a relational database server was deployed into the cloud. 4. those four VMs work with the VM hosting the database server. two servers were dedicated to payroll processing. a complex and time-consuming process. 4. Because of the size of the original (in-house) database.5 Portability Concerns: The payroll application runs on Fedora and Java 1. To avoid changing data structures and applications to use a cloud database. Instead of rewriting the application to use a cloud database service. The only API used was the S3 cloud storage API.1. The payroll application was deployed to four VMs that run simultaneously. That extracted information was transferred to a cloud storage service and then used by the database server. 4. The configuration of the payroll application was changed to use the VM hosting the database server. The payroll application used an SQL database for processing employee data. The database server retrieved data from a cloud storage system and constructed relational tables from it. so moving it to the cloud was relatively straightforward. The organization decided to see how practical it would be to run the payroll process in the cloud. so it will run without changes on any cloud provider's platform that supports Fedora. Finally. the cloud-based version is much more elastic.Cloud Computing Use Cases White Paper Version 4.3 Customer problem solved: In the cloud-based version of the application. changing the application to use a cloud database could be extremely 2 July 2010 34 . otherwise the application was not changed. extraction tools were used to select only the information necessary for payroll processing. The existing payroll system was architected as a distributed application.2 Customer scenario: In this scenario. that will be a significant advantage as the organization expands. The original application used a relational database. it was simply deployed to the virtual machine.1. As an added benefit.4 Requirements and Capabilities: The cloud services used were virtual machines and cloud storage (IaaS).1. Modifying the application to use a different cloud storage provider could be a problem if the other vendor doesn't support the specific S3 APIs used in the payroll process. The payroll application did not have to be modified. Finally. processing time for the payroll task was reduced by 80%.

The datastore maintains RDF graphs specific to the applications it is serving within silos managed on the server. All of the business logic resides on the client. Using this system. The REST architecture of the application allows HTTP’s built-in caching support to automatically propagate changes to the master datastore down to the client. If a client application does not need to access to certain fields or records. The client application maintains a local datastore that contains a subset of the most recent changes to the data. The datastore does not enforce any sort of schema other than an RDF graph. The solution to the problem was to build a custom client-side application. any number of applications can use the datastore without building a new code base for each one. 4. so they used a combination of Quickbooks and spreadsheets. The company had very specific requirements that no commonly available system addressed.2 Customer Scenario: Logistics and Project Management in the Cloud 4. In addition to the performance benefits of using a subset of the data.2 Customer scenario: A small construction company with approximately 20 administrative employees needed a way to manage their resources. this design simplifies security. although it does host an RDF-OWL ontology.Cloud Computing Use Cases White Paper Version 4.2.0 difficult.1 Applicable Use Case Scenario from Section 3: Enterprise to Cloud to End User 4. 2 July 2010 35 . The data set was small and easily processed using local resources. particularly if it involved moving to a cloud database that does not support the relational model.2. The client uses that ontology to validate data before displaying it to the user or sending it back to the GAE. optimize project scheduling and track job costs. Data for the application is served from a Google App Engine (GAE) datastore. This system was not elastic and was a huge waste of human resources. Security is implemented separately for each silo depending on the requirements of the application using a particular silo of data. Data operations are communicated with the datastore using an applicationspecific RESTful protocol over HTTP. Data was moved into the datastore on GAE from the locally hosted Quickbooks SQL server and the custom spreadsheets using a one-time data migration script that reconciled the data before uploading it to the GAE datastore. that portion of the datastore never leaves the server.

This is a significant difference from most datastores. will be virtualized to a public cloud. distributed. making future development and maintenance much simpler.2.1 Applicable Use Case Scenario from Section 3: Private Cloud 4. The combination of a RESTful API and the cloud datastore made the application more elastic than an application built around a traditional relational database. multi-dimensional sorted map that achieves elasticity by prioritizing denormalization over normalization. Although the original application infrastructure is still in use. 4. Some front office systems. will be virtualized and hosted in the private cloud. persistent.5 Portability Concerns: The application runs on the Google App Engine and its BigTable database. Porting the application to run on top of a more traditional datastore would require major changes to the application’s architecture. BigTable is a sparse. accounting and personnel management.2 Customer scenario: The ministries of the Japanese Government have thousands of servers across their infrastructures. a PaaS implementation that provides database support. cutting and pasting data by hand is no longer part of the process. and requires a fundamental rethinking of application development.3 Customer problem solved: Data was moved from an inefficient system of software and spreadsheet macros into a cloud-based system.2.4 Requirements and Capabilities: The cloud service used the Google App Engine. centralized infrastructure for hosting government applications. 4.0 4.3. In addition. 4. The resulting datastore can be used by a wide range of applications. removing a tedious task and eliminating a source of errors. such as electronic procurement. such as payroll.Cloud Computing Use Cases White Paper Version 4.3. The central government has announced a private “Kasumigaseki” cloud environment to provide a secure. the applications built on that infrastructure no longer rely on spreadsheets to analyze and manipulate the data.2. Existing back office systems.3 Customer Scenario: Central Government Services in the Cloud 4. but 2 July 2010 36 . Significant savings will come from the fact that maintaining the spreadsheets will no longer be necessary.

5 Portability Concerns: Because the government is building its own private cloud to host its own applications.4 Requirements and Capabilities: The cloud infrastructure will be built on a private network built by Japanese telecommunications companies. Wherever possible.3. 4. while other data will be stored locally. A secondary goal of the Kasumigaseki cloud is to provide a hybrid cloud environment. 4.4. The ultimate goal of the project is to reduce the total cost of ownership by eliminating redundant systems and the need for administrators in each ministry. 4. Because privacy and security are major concerns.3 Customer problem solved: The three problems solved by the hybrid cloud are reduced costs.3. In addition to the Kasumigaseki cloud. Each prefecture will have a private cloud and a connection to the Kasumigaseki hybrid cloud. portability is not a concern. 4. It is illegal for many types of personal data to be stored on a server outside of Japan. 2 July 2010 37 . each of which has its own servers and IT staff. Internal tasks and some data will be hosted in the prefecture’s private cloud.Cloud Computing Use Cases White Paper Version 4. reduced energy consumption and reduced IT staff. reduced energy consumption and reduced IT staff.1 Applicable Use Case Scenario from Section 3: Hybrid Cloud 4.2 Customer scenario: There are more than 1800 local governments across Japan.4 Customer Scenario: Local Government Services in a Hybrid Cloud 4.3 Customer problem solved: The three problems solved by the Kasumigaseki cloud are reduced costs.3.4. a private cloud is required.4. existing systems will be virtualized and hosted in the Kasumigaseki cloud. The government has no intention of moving its centralized applications and data from the private cloud to a public one. 4. the Japanese central government has decided to group local governments at the prefecture level.0 that is outside the scope of this project.

such as extra-solar planets and failed stars called brown dwarfs. 2 July 2010 38 . distances.html for an overview of the project. A framework for distributed computing (developed in house) is used for job execution and data processing. so moving applications and data into the Kasumigaseki cloud is not an option.1 Applicable Use Case Scenario from Section 3: End User to Cloud 4. 4. and changes in brightness. The government has no intention of moving its centralized applications and data from the private cloud to a public one.5. retrieves the data. federation of applications and data inside the hybrid cloud is crucial.0 4. For processing. The process runs a number of iterations over the data until it converges. 4. movements. It is expected to discover hundreds of thousands of new celestial objects.Cloud Computing Use Cases White Paper Version 4. This mission will collect a large amount of data that must be analyzed. each working node gets a job description from the database.5. Japanese law prevents some types of data from being stored outside the local government’s servers. processes it and sends the results to intermediate servers. precisely charting their positions. The prototype system contains the scientific data and a whiteboard used to publish compute jobs.4 Requirements and Capabilities: Privacy and security are crucial in this scenario. The goals were to determine the technical and financial aspects of using cloud computing to process massive datasets. Because some processing will be on the local government’s infrastructure.4. 8 See http://www. The framework is configured to run AGIS (Astrometric Global Iterative Solution). portability is not a concern because the government is building its own private cloud.2 Customer Scenario: Gaia8 is a mission of the European Space Agency that will conduct a survey of one billion stars in our galaxy.5 Customer Scenario: Astronomic Data Processing 4. The ESA decided to prototype a cloud-based system to analyze the data.int/esaSC/120377_index_0_m. The intermediate servers update the data for the following iteration.5 Portability Concerns: As with the previous customer scenario.4. It will monitor each of its target stars about 70 times over a five-year period.esa.

2 July 2010 39 . To evaluate the elasticity of a cloud-based solution. not a cloud database service. The prototype went through 24 iterations of 100 minutes each. The prototype allowed the organization to find and solve performance and elasticity problems before going to production. Storage of the datasets will be cloud-based as well. The database is a traditional database running inside a VM. five cloud-based storage volumes (100GB each) were attached to the database server. it is a full relational database. The portability concern for this application would be the ability to migrate those VM images to another provider without having to rebuild or reconfigure the images.200 hours. 4.Cloud Computing Use Cases White Paper Version 4. 100 million primary stars will be analyzed along with 6 years of data. a couple of performance problems with SQL queries and lock contention at the database were detected and solved. For the full billion-star project. In the second test. These problems could not have been detected with the current in-house system. 4.5. With each VM running 12 threads.0 The prototype evaluated 5 years of data for 2 million stars. so the actual savings will be even greater. equivalent to running a grid of 20 VMs for 40 hours. That cost estimate does not include the additional electricity or system administration costs of an in-house solution.5.5 Portability Concerns: All of the VMs are running standard operating systems and none of the software used in the project is cloud-specific.5. there were 1440 processes working in parallel. which will require running the 20 VM cluster for 16. the prototype ran a second test with 120 high CPU extra large VMs. 4.3 Customer Problem Solved: The estimated cost for the cloud-based solution is less than half the cost of an inhouse solution. a small fraction of the total data that must be processed in the actual project.4 Requirements and Capabilities: The prototype used VMs for the AGIS software and the database. For storage.

 Caching: A programming toolkit that supports caching for cloud resources would boost the performance of cloud applications. APIs should support writing logs. so developers writing cloud applications will choose a cloud database provider based on the needs of each application. monitor and pause processing jobs. A federated identity management system can produce credentials to allow a given user to access a particular service even if the service and user have no knowledge of each other. particularly in a cloud environment. APIs for identity management should cache and delete credentials as appropriate.0 5 Developer Requirements The requirements for developers cut across all of the use cases and customer scenarios. The API should allow developers to start. many are neither).Point-to-Point: Developers need an API to post messages to a queue and to consume those messages. this requires a way to authenticate a given user's credentials.  Identity Management: Developers need a way to manage identities. In the simplest case. The infrastructure of a cloud is 2 July 2010 40 .Cloud Computing Use Cases White Paper Version 4.  Database: Developers need a way to access cloud databases. Cloud databases vary widely in their design (some are schema-driven and relational.  Centralized Logging: Logging is a common requirement for all developers.  Raw Compute / Job Processing: Developers need an API to work with large processing jobs such as Hadoop-style data mining. The API should allow developers to post messages to a topic and retrieve messages from a topic. and might need an API to specifically put an object or resource into the cache. creating logs and opening and closing log files. An API must also allow the developer to peek the message (examine the message’s contents without consuming it). regardless of their role or the type of application they are writing.  Messaging . To simplify the discussion.  Messaging . Developers need an API to flush the cache. In cases where an application works with multiple data sources and applications.  Session Management: The ability to manage user sessions is crucial. stop.Pub-Sub: Developers need an API to work with topics in a message queuing system. a developer needs a way to federate that user's identities. The database API must support the basic CRUD operations. all developer requirements are listed here. examining entries.

2 with the requirements of developers.  Storage: Developers need a way to access cloud storage services.Cloud Computing Use Cases White Paper Version 4.4. The session API must make it easy to access or manipulate the current state of the user session.  Service Discovery: Developers need a way to discover which cloud services are available. Image and Infrastructure Management Ordinary Programming Deployment Developer Requirement Caching Centralized Logging Database Identity Management Messaging – Point-to-Point Messaging – Pub-Sub Raw Compute / Job Processing Service Discovery                 2 July 2010 Internal Interfaces Cloud Services       41 . With such an API.0 redundant and resilient in the face of machine failures. Cloud services should be searchable by type of service. developers can write applications that interrogate cloud services and select the one that best meets the application’s SLA criteria. The API must provide the ability to store and retrieve both data and metadata. The following table cross-references the five different categories of APIs discussed in Section 2. so sessions must be maintained even when a particular cloud node goes down.  SLAs: Developers using service discovery need an automated way to determine the policies of the services their code discovers. with the API providing additional function appropriate to each service type.

0 Image and Infrastructure Management Ordinary Programming Deployment Developer Requirement Session Management SLAs Storage          2 July 2010 Internal Interfaces Cloud Services   42 .Cloud Computing Use Cases White Paper Version 4.

Laws and regulations will determine security requirements that take priority over functional requirements. access control is just as important.  Security Controls: Although a given consumer might need all of these security controls. An important point to keep in mind is that the cloud does not introduce any new security threats or issues. cloud computing as a whole can be considered the ideal use case to highlight the need for a consistent. in the cloud or elsewhere. Cloud providers should deliver these patterns through existing security standards. The most significant difference when considering security from a cloud perspective is the enterprise’s loss of control. 6. platform and application of security is under the direct control of the cloud provider. (These have been mentioned earlier in this paper. is a crucial topic that could fill any number of pages.Cloud Computing Use Cases White Paper Version 4. but they must be addressed. transparent.) 2 July 2010 43 .1 Regulations Beyond all of the technical issues in using cloud computing is the harsh reality of regulations. With an in-house application. but the infrastructure. With a cloud-based application. This section discusses security topics in the following order:  Regulations: Regulations are not technical issues. but they warrant more discussion within the context of security. several federation patterns are needed.  Security Federation Patterns: To implement these security controls. as opposed to any particular technical challenge. To put security in perspective. controlling access to sensitive data and applications is crucial. This information sets the tone for the discussion of the user scenarios in the next section. As companies move or build solutions in the cloud. consumers should be wary of any cloud provider that makes security-related claims and reassurances without an infrastructure capable of delivering all of them. The purpose here is to highlight the security issues that architects and developers should consider as they move to the cloud.0 6 Security Scenarios Security. standards-based security framework regardless of cloud deployment model. having this consistent security model is vital to simplify development and to avoid vendor lock-in and preserve their IT investments.

This includes being able to account for any physical. This includes employing standards-based cryptographic functions and services to support information security at rest and in motion. Security Control Asset Management Cryptography: Key and Certificate Management Data / Storage Security Description It must be possible to manage all of the hardware. It must be possible to store data in an encrypted format. although new laws and regulations will be created on an ongoing basis. some consumers will need their data to be stored separately from other consumers' data. There are often stiff penalties for organizations (and. If a virtual machine is running in the cloud. they represent best practices. 2 July 2010 44 . The following section describes several use cases and the security requirements for each. In addition.or network-based access of an asset for audit and compliance purposes.2 Security Controls A number of controls are necessary to adequately secure any system.Cloud Computing Use Cases White Paper Version 4. Any organization storing sensitive data in the cloud must be able to prove that their cloud provider never stores that data on a physical server outside a particular geographic area. The CIO’s office must manage these changes and be alert for new laws and regulations as they emerge. In addition to government agencies. The discussion of the use cases relates their requirements to the controls defined here. A new law might require an organization to spend their resources changing an application’s infrastructure instead of adding features to it. Following these laws and regulations will take precedence over all other requirements. While those regulations might not be required by law. network and software assets (physical or virtual) that make up the cloud infrastructure. governments around the world are concerned about the use of cloud computing. 6. many trade and industry groups create regulations as well.0 For a variety of reasons. Any secure system needs an infrastructure for employing and managing cryptographic keys and certificates. their executives) that violate those laws. Similar concerns apply to applications running in the cloud. can an application running on that VM access sensitive data? This is a gray area that many countries have not addressed. in some cases. Many countries have strict privacy laws that prohibit certain data from being stored on a physical machine located outside that country.

especially system failures and security breaches. There must be an automated way to manage and analyze security control flows and processes in support of security compliance audits. It must be possible to configure.0 Security Control Endpoint Security Description Consumers must be able to secure the endpoints to their cloud resources.org/committees/kmip/) 2 July 2010 45 . It must be possible to define the identity. roles. Access Control and Attributes Network Security Security Policies Service Automation Workload and Service Management Here are some standards that can be applied to these controls: Security Control Cryptography: Key and Certificate Management Relevant Standard KMIP. entitlements and any other attributes of individuals and services in a consistent.Cloud Computing Use Cases White Paper Version 4. and enforce security policies in support of access control. deploy and monitor services in accordance with defined security policies and customer licensing agreements. the OASIS Key Management Interoperability Protocol (http://www.oasis-open. Cloud providers cause significant damage to their reputations when they fail to report events in a timely manner. resource allocation and any other decisions in a consistent. It must be possible to secure network traffic at the switch. The IP stack itself should be secure as well. Event Auditing and Reporting Identity. This also includes reporting any events that violate any security policies or customer licensing agreements. machinereadable way. This includes the ability to restrict endpoints by network protocol and device type. Roles. Access to events includes the ability to learn about past events and reporting of new events as they occur. router and packet level. It must be possible to define policies. machine-readable way in order to effectively implement access control and enforce security policy against cloud-based resources. resolve. The method for defining policies should be robust enough that SLAs and licenses can be enforced automatically. Consumers must be able to access data about events that happen in the cloud.

so the many assets. Access to resources can be controlled by more than one factor.org/) X. the OASIS eXtensible Access Control Markup Language (http://www.int/rec/T-REC-X. and then using those credentials to secure messages and create signed security tokens (typically SAML). access 2 July 2010 46 . That authentication authority is capable of exchanging credentials (typically X. Cloud computing itself is a federation of resources. etc. part of the ITU Public Key and Attribute Certificate Frameworks Recommendations (http://www.itu. the OASIS Security Assertion Markup Language (http://saml. developed by the IEEE Security in Storage Working Group (https://siswg. For example.  Access Management: The ability to write policies (typically in XACML) that examine security tokens to manage access to cloud resources. even though the service provider has no knowledge of the user.net/) SAML.509 certificates). Service providers that trust the identity provider can use that token to grant appropriate access to the user.509 Certificates.3 Security Federation Patterns Federation is the ability of multiple independent resources to act like a single resource. identities.0 Security Control Data / Storage Security Identity.) and returns a signed security token that identifies that user.Cloud Computing Use Cases White Paper Version 4.  Identity Management: The ability to define an identity provider that accepts a user’s credentials (a user ID and password. Federated trust is the foundation upon which all the other secure federation patterns are based. Roles.509) XACML. The requirements from the previous section are implemented via the following federation patterns:  Trust: The ability for two parties to define a trust relationship with an authentication authority.org/committees/provision/) 6. Roles.oasis-open.xml.oasis-open. configurations and other details of a cloud computing solution must be federated to make cloud computing practical. Access Control and Attributes Identity. a certificate. the OASIS Service Provisioning Markup Language (http://www.org/committees/xacml/) SPML. Access Control and Attributes Security Policies Workload and Service Management Relevant Standard IEEE P1619.

 Audit and Compliance: The ability to collect audit and compliance data spread across multiple domains. This data can include access policies and licensing information across multiple domains. Federated single sign-off is part of this pattern as well. Federated audits are necessary to ensure and document compliance with SLAs and regulatory requirements. applications and virtual machines.  Single Sign-On / Sign-Off: The ability to federate logins based on credentials from a trusted authority. Because existing best practices for security apply to the cloud. but only across certain protocols and only at certain times of the day. 2 July 2010 47 . Given an authenticated user with a particular role. providers should use existing standards to deliver these federation patterns.Cloud Computing Use Cases White Paper Version 4. The Single Sign-On pattern is enabled by the Identity Management pattern. federated single sign-on allows a user to login to one application and access other applications that trust the same authority.  Configuration Management: The ability to federate configuration data for services.0 to a resource could be limited to users in a particular role. including hybrid clouds. in many situations it will be vital that a user logging out of one application is logged out of all the others.

2 Customer Problem Solved: Using a public cloud allows the company to handle workloads that could be an order of magnitude greater than usual. 7. powering and cooling enough extra systems to handle the highest imaginable work load is not cost-efficient.3 Requirements and Controls: The requirements and associated controls for this use case are: Requirement Access to videos limited to particular roles Controls  Identity.1 Computing Power in the Cloud 7. deployment models. A hurricane is projected to strike the Gulf Coast region of the United States.1. The company’s decision is to use a public cloud provider to deliver virtual machines to handle the expected demand. 7. In addition. Finally. yet they can allocate additional computing power automatically as needed. Access Control and Attributes  Asset Management  Network Security 2 July 2010 48 . Roles. The company has the in-house capability to handle its day-to-day operations. the company must securely transmit any data created by cloud-based instances of the application back inside the corporate firewall. we look at real-world scenarios to illustrate cloud security in action. maintaining.1. the cloud provider must ensure that no traces of the application or its data remain whenever a virtual machine is shut down. patterns and roles. likely causing massive property damage. which will in turn create an enormous load on the corporate IT infrastructure. The scenarios cover a range of application types.0 7 Security Use Case Scenarios This section describes customer experiences using cloud computing while meeting security requirements. limiting access to only authorized agents of the company. The company must control access between the enterprise system and the virtual machines hosted by the cloud provider. Buying.Cloud Computing Use Cases White Paper Version 4. This will create a huge spike in claims.1.1 Customer Scenario: An insurance company has a claims application used to capture data about their policy holders and any property damage they suffer. 7. With requirements and patterns defined.

Access Management.0 interface. 7.1. as the company moves from a traditional interface to a Web 2.5 Roles Claims adjusters.0 interface has far more interactions with the server than a static Web page. using cloud-hosted development tools eliminates the need to install. as large builds can use the scalability of the cloud provider’s infrastructure.2. the extra burden on the server is an unknown.2. configure and manage tools on each developer’s machine.4 Federation Patterns: Trust. but does not want to burden its IT staff and existing resources.1 Customer Scenario: An online retailer needs to develop a new Web 2.2 Customer Problem Solved: From a development perspective.0 storefront application. In addition.Cloud Computing Use Cases White Paper Version 4. If the testing team wants to see how the application performs when requests are coming in from 500 different machines per second. The company chooses a cloud provider to deliver a cloud-based development environment with hosted developer tooling and a source code repository. Configuration Management 7.2 Cloud-based Development and Testing 7.1. insurance agents.0 Requirement Controls All traces of the application or data must  Workload and Service Management be deleted when a VM is shut down 7. auditors 7. Stress testing the new application in the cloud is much easier. Another cloud provider is chosen to provide a testing environment so that the new application can interact with many different types of machines and huge workloads. Product builds can be much faster. From a testing perspective. therefore testing the new application in the cloud allows the testing team to measure the impact of the new interface. at the cloud provider’s site. Choosing two providers to handle cloud-based development and testing means federation is crucial. and all developers are automatically moved to the same version of the tools. A Web 2. cloud-based builds can retrieve the latest source code from the cloud-based repository. Updates to the tools are installed once. the cloud allows them to start 500 virtual 49 2 July 2010 .

Cloud Computing Use Cases White Paper Version 4. etc. it is cost-effective to do so in the cloud.2. If the testing team wants to run the same test with 1.).2.2.3 Storage in the Cloud 7. Single Sign-On. Roles. protocols. administrators. Configuration Management 7.000 machines or 10. 7.4 Federation Patterns: Trust. versions.1 Customer scenario: A financial investment company is launching new investment products to its agents and affiliates.000 machines.0 machines and run those tests. With different VM images.3. Roles. Identity Management. Access Management. Access Control and Attributes  Service Automation Single sign-on across development and testing clouds Controlled access to source code and test plans Builds and tests must start and shut down VMs automatically Builds and tests must report statistics on  Event Auditing and Reporting VM usage and performance 7. building the infrastructure in-house is not.5 Roles: Developers.3 Requirements and Controls: The requirements and associated controls for this use case are: Requirement Controls Developer tools installed and maintained  Asset Management in one central location All traces of the application or data must  Workload and Service Management be deleted when a VM is shut down  Cryptography  Endpoint Security  Identity. Access Control and Attributes  Network Security  Asset Management  Identity. Audit and Compliance. it is straightforward to test the application with many different types of machines (operating systems. A number of videos have been created to teach the 2 July 2010 50 . testers. auditors 7.

The cloud solution must control the videos with an auditable access control mechanism that enforces the company’s security policies.3 Requirements and Controls: The requirements and associated controls for this use case are: Requirement Controls  Identity.0 company’s agents and affiliates about the benefits and features of the new products. access to those videos needs to be tightly controlled. confidential during the quiet period before the launch of the product. including the videos. Access Management. Audit and Compliance 2 July 2010 51 . A public cloud storage provider that cannot guarantee compliance is not an option. 7. However.2 Customer Problem Solved: Using a public cloud storage service allows the customer to manage massive data files and bandwidth demands without increasing the physical resources of their data center. For competitive reasons. government regulations and organizational requirements mean security is vital. Access Control and Attributes  Asset Management  Network Security  Policies  Cryptography  Data / Storage Security     Cryptography Data / Storage Security Endpoint Security Network Security Access to the application limited to particular roles Data stored in the cloud must be secured Data stored in the cloud must be transferred back inside the company’s firewall 7.3. However. Identity Management. An even stronger constraint is that regulations require the company to keep product details. The company’s decision is to use a public cloud storage provider to scale the secure hosting and streaming of the videos. no matter what performance. The videos are very large and need to be available on-demand. Roles.Cloud Computing Use Cases White Paper Version 4. so storing them in the cloud lessens the demands on the corporate infrastructure. only certified company agents should be able to view the videos.4 Federation Patterns: Trust. 7.3. price or scalability they provide.3.

company agents. regulatory authorities 7.5 Roles: Video producers. Roles.Cloud Computing Use Cases White Paper Version 4. Access Control and Attributes Network Security Policies Service Automation Workload and Service Management Storage in the Cloud                   2 July 2010 52 .0 7.3. auditors. company affiliates.4 Cross-Reference: Security Controls and Customer Scenarios The following table summarizes the relationships between the security controls and the customer scenarios: Computing Power in the Cloud Cloud-based Development and Testing Security Control Asset Management Cryptography: Key and Certificate Management Data / Storage Security Endpoint Security Event Auditing and Reporting Identity.

Cloud Computing Use Cases White Paper

Version 4.0

7.5 Cross-Reference: Security Federation Patterns and Customer Scenarios
The following table summarizes the relationships between the security federation patterns and the customer scenarios: Computing Power in the Cloud Cloud-based Development and Testing

Security Federation Pattern Trust Identity Management Access Management Single Sign-On Audit and Compliance Configuration Management

Storage in the Cloud

 

     

   

2 July 2010

53

Cloud Computing Use Cases White Paper

Version 4.0

8 Service Level Agreements (SLAs)
The relationship between the cloud provider and the cloud consumer must be described with a Service Level Agreement. Because cloud consumers trust cloud providers to deliver some of their infrastructure services, it is vital to define those services, how they are delivered and how they are used. An SLA is the foundation of the consumer's trust in the provider. A well-written SLA codifies the provider's reputation. In addition to the prose that defines the relationship between the consumer and provider, an SLA contains Service Level Objectives (SLOs) that define objectively measurable conditions for the service. The consumer must weigh the terms of the SLA and its SLOs against the goals of their business to select a cloud provider. It is crucial that the consumer of cloud services fully understand all the terms of the provider's SLA, and that the consumer consider the needs of their organization before signing any agreement.

8.1 What is an SLA?
An SLA defines the interaction between a cloud service provider and a cloud service consumer. An SLA contains several things:  A set of services the provider will deliver  A complete, specific definition of each service  The responsibilities of the provider and the consumer  A set of metrics to determine whether the provider is delivering the service as promised  An auditing mechanism to monitor the service  The remedies available to the consumer and provider if the terms of the SLA are not met  How the SLA will change over time The marketplace features two types of SLAs: Off-the-shelf agreements and negotiated agreements between a provider and consumer to meet that consumer's specific needs. It is unlikely that any consumer with critical data and applications will be able to use the first type. Therefore the consumer's first step

2 July 2010

54

Cloud Computing Use Cases White Paper

Version 4.0

in approaching an SLA (and the cloud in general) is to determine how critical their data and applications are. Most public cloud services offer a non-negotiable SLA. With these providers, a consumer whose requirements aren't met has two remedies: 1. Accept a credit towards next month's bill (after paying this month's bill in full), or 2. Stop using the service. Clearly an SLA with these terms is unacceptable for any mission-critical applications or data. On the other hand, an SLA with these terms will be far less expensive than a cloud service provided under a negotiated SLA.

8.2 Service Level Objectives
An SLO defines a characteristic of a service in precise, measurable terms. Here are some sample SLOs:  The system should never have more than 10 pending requests.  Throughput for a request should be less than 3 seconds.  Data streaming for a read request should begin within 2 seconds.  At least five instances of a VM should be available 99.99999% of the time, with at least one instance available in each of a provider's three data centers. Obviously different Service Level Objectives will apply to different use cases, applications and types of data. SLOs can also include an urgency rating to indicate the relative importance of different SLOs. A consumer could use an urgency rating to indicate that availability is more important than response time if the cloud provider cannot deliver both SLOs. Different roles also affect the SLOs that apply. For example, consider an application written by a cloud consumer, hosted by a cloud provider and accessed by an end user. If the application and its data are hosted by the same cloud provider, chances are the application can access that data without leaving the provider's data center. The cloud consumer will expect very fast response times whenever the application accesses its data. On the other hand, the consumer will have lower expectations of response times whenever an end user accesses the application across the Web.

2 July 2010

55

a consumer might notice that the CPU load on a particular VM is above 90%. if giving one consumer more resources would make it impossible to meet the terms of another consumer's SLA. A consumer must select providers and services based on the goals of the organization. Getting all parts of an organization to agree on those goals will involve some groups accepting budget cuts.0 8. For example. As with the provider. 8.4. if the consumer's SLA says that the first 100 VMs are priced at one rate. However. the consumer might decide not to create another VM and incur higher charges. there are a number of factors they should consider. 8. Service Level Management is how that performance information is gathered and handled. for that matter) until the business level objectives have been defined. Despite these difficulties. Service Level Management helps consumers make (and possibly automate) decisions about the way they use cloud services. A cloud consumer uses Service Level Management to make decisions about how it uses cloud services.Cloud Computing Use Cases White Paper Version 4. In response. 2 July 2010 56 . For example. A cloud provider uses Service Level Management to make decisions about its infrastructure. Measurements of the service are based on the Service Level Objectives in the SLA. the provider might decide to keep one customer happy at the expense of another. the consumer might start another VM. However. a provider might notice that throughput for a particular service is barely meeting the consumer's requirements. the organization cannot make the most of cloud computing (or any technology.3 Service Level Management It is impossible to know whether the terms of the SLA are being met without monitoring and measuring the performance of the service. some groups losing control of their infrastructure and other difficult choices. This is an aspect of cloud computing in which the hardest problems are organizational politics rather than technical issues. The provider could respond to that situation by reallocating bandwidth or bringing more physical hardware online. with more VMs priced at a higher rate. Defining exactly what services will be used is worthless unless the organization has defined why it will use the services in the first place.1 Business Level Objectives Debating the terms of an SLA is meaningless if the organization has not defined its business level objectives. The goal of Service Level Management is to help providers make intelligent decisions based on its business objectives and technical realities.4 Considerations for SLAs As consumers are deciding what terms they need in an SLA.

but none of that matters if the cloud provider itself does not have adequate continuity and disaster recovery procedures. 8. consumers should understand how and when their providers will do maintenance tasks. Some consumers store copies of valuable data in multiple clouds for backup.3 Business Continuity and Disaster Recovery Many consumers use the cloud for business continuity.Cloud Computing Use Cases White Paper Version 4. On the other hand.0 Consumers should know why they are using cloud computing before they decide how to use cloud computing. they should be clear about the consumer's responsibilities as well. Consumers should ensure their cloud providers have adequate protection in case of a disaster.4 System Redundancy Many cloud providers deliver their services via massively redundant systems. Will their services be unavailable during that time? Will their services be available. restrictions on the type of data that can be stored. 8.4. The cloud can be an invaluable resource to keep an organization running when in-house systems are down. 8. Consumer responsibilities might include limits on system usage. will the consumer have a chance to test their applications against the updated service? Note that maintenance can affect any type of cloud offering and that it applies to hardware as well as software.4. 2 July 2010 57 . However. The balance of responsibilities between providers and consumers will vary according to the type of service. Other consumers use cloudbursting when in-house data centers are unable to handle processing loads. Consumers moving data and applications that must be constantly available should consider the redundancy of their provider's systems. Those systems are designed so that even if hard drives or network connections or servers fail. 8. a VM that contains licensed software and works with sensitive data places many more responsibilities on the consumer who builds and manages it.4.4. or having a valid license for any software used on the provider's systems. consumers will not experience any outages. freeing consumers from having to do that themselves. a cloud provider bears most of the responsibilities for Software as a Service. For example. but with much lower throughput? If there is a chance the maintenance will affect the consumer's applications.5 Maintenance Providers handle the maintenance of their infrastructure.2 Responsibilities of the Provider and Consumer Although SLAs are commonly thought of as defining the provider's responsibilities.

Even if law enforcement targets the data and applications associated with a particular consumer.8 Failure of the Provider Any cloud provider has the potential to either go out of business or be acquired by another company.4. If a cloud provider cannot guarantee that a consumer's data will be stored in certain locations only. For example. the multi-tenant nature of cloud computing makes it likely that other consumers will be affected.9 Jurisdiction Consumers should understand the laws that apply to any cloud providers they consider.7 Seizure of Data There have been a few well-publicized instances of law enforcement officials seizing the assets of a hosting company. many countries prohibit storing personal information about its citizens on any machine outside its borders. a cloud provider could be based in a country that reserves the right to monitor any data or applications using that cloud provider's services. 8. 8.10 Cloud Brokers and Resellers If a cloud provider is actually a broker or reseller for another cloud provider.4. In addition. this might not be acceptable. Consumers should consider the financial health of their provider and make contingency plans if the provider were to shut its doors. Consumers should also consider using a third party to back up their data and applications.4. the terms of the SLA should clarify any questions of responsibility or liability if anything goes wrong at the broker. Although there are limits to what an SLA can cover. consumers should consider the laws that apply to the provider.0 8.Cloud Computing Use Cases White Paper Version 4. reseller or provider facilities. the provider's policies on access to the consumer's data and applications if the consumer's account is delinquent or in dispute should be made clear.4. If a cloud service provider promises to enforce data location regulations. the consumer must be able to audit the provider to prove that regulations are being followed. 8.4. 8. 2 July 2010 58 . For example. the consumer cannot use that provider's services.6 Location of Data The physical location of many types of data is restricted. Given the nature of the consumer's data and applications.

Cloud providers must be able to prove they are compliant with these policies.5. If a cloud provider's hard drive fails. On a similar note.5 SLA requirements 8. In addition. In turn. the cloud provider must be able to prove their compliance. an SLA should make it clear how the cloud provider isolates data and applications in a multi-tenant environment. the platters of that disk should be zeroed out before the drive is disposed or recycled.4 Data Retention and Deletion Many organizations have legal requirements that data must be kept for a certain period of time.5.5 Hardware Erasure and Destruction A common source of data leaks is the improper disposal of hardware.5. A cloud consumer must understand their security requirements and what controls and federation patterns are necessary to meet those requirements. Some of those are laws (HIPAA for medical records in the United States). Some organizations also require that data be deleted after a certain period of time.5. it is important that the data be encrypted while it is in motion and while it is at rest.6 Regulatory Compliance Many types of data and applications are subject to regulations.5. while others are industry-specific (PCI DSS for retailers who accept credit cards). a cloud provider must understand what they must deliver to the consumer to enable the appropriate controls and federation patterns. The security-related aspects of an SLA should be written with the security controls and federation patterns from Section 6 in mind.5.1 Security Security as a general requirement is discussed in detail in Sections 6 and 7 of this paper. 2 July 2010 59 . retention and deletion. 8. 8. The details of the encryption algorithms and access control policies should be specified in the SLA. many cloud providers offer the added protection of zeroing out memory space after a consumer powers off a VM. 8. 8.3 Privacy Basic privacy concerns are addressed by requirements such as data encryption.0 8.2 Data Encryption If a consumer is storing vital data in the cloud. If regulations must be enforced.Cloud Computing Use Cases White Paper Version 4. 8.

A set of industry-defined terms for different key performance indicators would make it much easier to compare SLAs in particular (and cloud services in general). Because the consumer is liable for any breaches that occur. It is in the provider's interest to define uptime in the broadest possible terms. Often that definition is specific to a provider's architecture.5. 8. while consumers could be tempted to blame the provider for any system problems that occur. 8. If a provider has a data center on six continents.0 8. the consumer bears the burden of proving that the provider failed to live up to the terms of the SLA.8 Certification There are many different certifications that apply to certain types of data and applications.10 Monitoring If a failure to meet the terms of an SLA has financial or legal consequences. does uptime refer to a particular data center or any data center? If the only available data center is on another continent.5. providers must be proactive in notifying consumers when the terms of the SLA are breached. For example.5.7 Transparency Under the SLAs of some cloud providers.9 Terminology for key performance indicators The term uptime can be defined in many ways. This includes infrastructure issues such as outages and performance problems as well as security incidents. This eliminates the conflicts of interest that might occur if providers report outages at their sole discretion or if consumers are responsible for proving that an outage occurred. A provider's service might be down for hours.Cloud Computing Use Cases White Paper Version 4. 8.5.11 Auditability Many consumer requirements include adherence to legal regulations or industry standards. To make matters worse. For critical data and applications. This makes it difficult to compare cloud services. consumer might have the requirement that their cloud provider be ISO 27001 certified.5. that uptime is unlikely to be acceptable. but consumers who are unable to prove that downtime are not eligible for any sort of compensation. the question of who should monitor the performance of the provider (and whether the consumer meets its responsibilities as well) becomes crucial. The best solution to this problem is a neutral third-party organization that monitors the performance of the provider. 8. it is vital 60 2 July 2010 . other cloud providers will use definitions specific to their architectures. The provider would be responsible for proving their certification and keeping it up-to-date.

but the most secure provider 2 July 2010 61 .12 Metrics Monitoring and auditing require something tangible that can be monitored as it happens and audited after the fact. An SLA should make it clear how and when those audits take place. with limits (the maximum amount of storage or bandwidth. 8.Cloud Computing Use Cases White Paper Version 4. One of the basic characteristics of cloud computing is on-demand self-service. 8. Although listing all metrics it is impossible. For example.5.5. Cloud consumers will have an endless variety of metrics depending on the nature of their applications and data. This refers to the human interactions required when something goes wrong with the on-demand. the consumer's policy might state that the broker should use the cheapest possible provider for some tasks. self-service aspects of the cloud. Because audits are disruptive and expensive. The metrics of an SLA must be objectively and unambiguously defined. The broker could select a cloud provider based on business criteria defined by the consumer. the provider will most likely place limits and charges on them. for example) clearly stated  Linearity – How a system performs as the load increases  Agility – How quickly the provider responds as the consumer's resource load scales up and down  Automation – What percentage of requests to the provider are handled without any human interaction  Customer service response times – How quickly the provider responds to a service request.0 that the consumer be able to audit the provider's systems and procedures. some of the most common are:  Throughput – How quickly the service responds  Reliability – How often the service is available  Load balancing – When elasticity kicks in (new VMs are booted or terminated.13 Machine-Readable SLAs A machine-readable language for SLAs would enable an automated cloud broker that could select a cloud provider dynamically. an automated cloud broker would extend this characteristic by selecting the cloud provider on demand as well. for example)  Durability – How likely the data is to be lost  Elasticity – The ability for a given resource to grow infinitely.

Typical guarantees will cover how many requests the consumer can make. If a VM and a cloud database are in the same data center. On the other hand. but if the network connection between them fails. The service is unavailable when the first cloud provider's systems go down. These situations must be rare. To sum up. five nines reliability means the service is available 99. If a cloud provider offers a service built on a second cloud provider's storage service and a third cloud provider's database service. communication between the VM and the database doesn't require network access. the fact remains that there will always be problems that can only be resolved with human interaction. as the number of cloud providers increases. Both cloud providers cloud can be up and running with healthy systems.Cloud Computing Use Cases White Paper Version 4. As an example. (It loses even more meaning if the cloud provider gets to decide whether an incident constitutes an outage. 2 July 2010 62 .0 for others. Although substantial marketplace demand for this requirement will take some time to develop. a common metric bandied about is the number of “nines” a provider delivers.99999% of the time. any work on standardizing SLAs should be done with this in mind. which translates to total system outages of roughly 5 minutes out of every 12 months. whether directly or indirectly. but each additional provider makes the system less reliable. the available bandwidth between the VM and the database affects the performance and reliability of the overall system. Finally. but many SLAs will include guarantees about the provider's responsiveness to requests for support. it is important to consider that many cloud offerings are built on top of other cloud offerings. the reliability of the entire system is less than five nines. the more downtime the overall system will have.14 Human Interaction Although on-demand self-service is one of the basic characteristics of cloud computing. how much they will cost and how soon the provider will respond. One problem with this metric is that it quickly loses meaning without a clear definition of what an outage is. The more cloud providers involved. the service is equally unavailable when the second or third providers' systems have problems. the overall system is down.5. The ability to combine multiple infrastructures provides a great deal of flexibility and power. any consumer who needs to evaluate the reliability of a cloud service should know as much as possible about the cloud providers that deliver that service. 8. and all of those providers deliver five nines reliability. the number of outside factors increases as well. if the cloud database is delivered by another provider. 8.6 A note about reliability In discussions of reliability.) Beyond the nebulous nature of nines.

7 Cross-reference: SLA Requirements and Cloud Delivery Models The following table cross-references the three NIST delivery models from Section 2.0 8.8 Cross-reference: SLA Requirements and Use Case Scenarios At its best. Just as a given SLA doesn't meet the needs of all consumers. an SLA protects the interests of both the cloud consumer and cloud provider.1 with the SLA requirements listed here.1. Platform as a Service Infrastructure as a Service Software as a Service Requirement Data Encryption Privacy Data Retention and Deletion Hardware Erasure and Destruction Regulatory Compliance Transparency Certification Terminology for Key Performance Indicators Metrics Auditability Monitoring MachineReadable SLAs ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü 8. The 63 2 July 2010 . every requirement discussed here doesn't apply to all customer scenarios.Cloud Computing Use Cases White Paper Version 4.

Changing Cloud Vendors Private Cloud Enterprise to Cloud to End User Enterprise to Cloud Hybrid Cloud Enterprise to Cloud to Enterprise Requirement Data Encryption Privacy Data Retention and Deletion Hardware Erasure and Destruction Regulatory Compliance Transparency Certification Terminology for Key Performance Indicators Metrics Auditability Monitoring MachineReadable SLAs End User to Cloud ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü ü 2 July 2010 64 .Cloud Computing Use Cases White Paper Version 4.0 following table cross-references the seven use case scenarios from Section 3 with the SLA requirements listed here.

although the requirements for security will vary widely depending on the application and data types. As a result.0.Cloud Computing Use Cases White Paper Version 4. government agencies. and they will be central to any follow-on work. and lifecycle management. monitoring.  Security: Security in cloud computing is vital. As we go forward. but it quickly grew to include many other individuals around the world. deployment. proprietary solutions that lead to vendor lock-in.0 9 Conclusions and Recommendations Cloud computing builds on and complements many trends in the industry. data and application federation. This paper is validation that those principles work. three principles from the manifesto were crucial: 1) users should work together. governance.  Cloud Management: Cloud computing is not feasible without service management. The initial group consisted of supporters from the Open Cloud Manifesto. federated identity. Data Formats and APIs: Virtual machines. SOA and Web 2. The community included representatives from large and small companies. data and applications created for one cloud provider should run on another cloud provider without changes. This paper was created by an open Web community of more than 1400 participants. It must be possible for consumers to implement any of the use cases outlined here without resorting to closed. metering. 2) activities to keep the cloud open should be customer driven and 3) existing standards should be used wherever possible. The developer requirements discussed here can be grouped into two broad categories: 2 July 2010 65 . consultants and vendors.  Location awareness: A way of identifying the location of the physical machine hosting the cloud infrastructure is an absolute requirement for many government regulations. The use cases described here demonstrate the following general requirements for consumers:  Common VM Formats. SLAs and benchmarks. including virtualization. leverage standards work already in progress. we will work together as a community to specify the existing standards that meet customer needs. standards already exist for many of the requirements outlined in this paper. and identify what is needed to fill in the gaps not addressed by existing standards. These requirements are defined in Section 3. As the paper was developed.

 Support APIs: The API requirements for caching. Cloud computing does not introduce any security issues that have not already been raised for general IT security. Throughout all aspects of cloud computing. service discovery. messaging (both point-to-point and publish-subscribe). where existing standards meet requirements. 2 July 2010 66 . Where existing standards do not meet requirements. This communitywritten paper is meant to be the reference for establishing a truly open cloud computing environment. logging. and that the consumer consider the needs of their organization before signing any agreement. This loss of control emphasizes the need for transparency from cloud providers. federation patterns and standards that cloud providers must deliver. we must define and implement the standards needed to meet them. As organizations use cloud services. raw computing power and storage all relate directly to cloud services. An SLA defines how the consumer will use the services and how the provider will deliver them. The discussion of security here covers the security controls.0  APIs for Services: The API requirements for databases. Security consistently raises the most questions as consumers look to move their data and applications to the cloud.Cloud Computing Use Cases White Paper Version 4. proprietary APIs that lead to vendor lock-in. identity management. we must ensure those standards are implemented pervasively and consistently. It is crucial that the consumer of cloud services fully understand all the terms of the provider's SLA. the responsibilities of both the consumer and the provider must be clearly defined in a Service Level Agreement. It must be possible for developers to implement any of the use cases outlined here without resorting to closed. session management and SLAs are necessary to use cloud services effectively. The concern in moving to the cloud is that implementing and enforcing security policies now involves a third party.

2.0 Summary of Changes Version 2.2. Developer Requirements.  Added a brief summary of Service Level Agreements to the Conclusions and Recommendations section. and Section 7. 30 June 2010:  Added Section 8. Version 3. 2 July 2010 67 . 30 October 2009:  Added Section 5. made minor changes to the discussion of Service Level Agreements in Section 3.  Updated the taxonomy discussion in Section 2.1.  Added Section 6. Version 4. Developers and their requirements were the main focus of the new content in Version 2. categories of APIs and developer roles were added as Section 2. Service Level Agreements (SLAs). 2 February 2010:  With security in mind.  Replaced the word “scalable” with “elastic” where appropriate.Cloud Computing Use Cases White Paper Version 4.  Added a brief summary of security to the Conclusions and Recommendations section.  Fixed a typo (from “basic the identity” to “the basic identity”) in Section 3.  Expanded the discussions of the requirements for consistent deployment and common VM formats to include the different ways cloud vendors attach storage to virtual machines.1 to reflect changes to the NIST definition of cloud computing.1. Security Scenarios.4.4. A discussion of the different levels of APIs. Security Use Case Scenarios.  Updated the customer scenario in Section 4.

Sign up to vote on this title
UsefulNot useful